Seasons
This is a forum or general chit-chat, small talk, a "hey, how ya doing?" and such. Or hell, get crazy deep on something. Whatever you like.
Posts 5,355 - 5,366 of 6,170
fuck u colonal
Haha easy there, you're barking up the wrong tree. Hey, maybe there are some bots around here that are members of your sexual persuasion? You might have a hard time with them, but where there's a will there's a way.
It's great to hear from you colonel720. I can't wait to see your ideas in action..I am still having great fun with Nick.
I'm gald you still have fun with nick... I have not had the time to play around with it as much as i was able to during development, but maybe one of these days i'll put a virtual machine on one of my servers and have nick running there 24/7, or make a web interface and let people chat with it over the web.
Posts 5,355 - 5,366 of 6,170
Hi everyone!!!
I must confess, I haven't logged on in almost a year. I've been very busy. But, I'm glad to see that everyone is still here, and that the forge is running on new hardware.
On the note of Artificial Intelligence, here are my latest ideas, for (perhaps Nick 3????)
I am envisioning a system that integrates neural adaptation and evolutionary algorithms. At the lowest level, MLP's (multi-layer perceptrons) will accept input data from cameras and microphones, or any other sensory input. The MLP's will adapt to recognize objects, and output to some form of SOM (self organizing map). the SOM will categorize the data into regions of similarity, which will be stored in memory, which would be a massive heirarchical tree of data. each "Region" of similar data categorized by the SOM will be treated as an object, or an instance of a more theoretical "class" (object oriented programmers are very familiar with that concept). The memory tree will consist of these classes which represent concepts formed from large amounts of input information. This way, after seeing the instances of the same patterns in its perceptions, it can combine all of those patterns into a generalization of them.
On a higher level, a recursive, evolving, self optimizing algoritm takes this conceptual data stored in the tree, and finds patterns and relationships between them. each time the algorithm loops through a data set, it can build on the higer level connections made during its last loop, and finds information and pattern abstractions that could not be found in any previous loops.
At the moment, I am working on a multithreaded neural network model. Lots more work to do.
contrary to earlier projects, I no longer believe that text should be primary sensory input for an AI. Rather, the ability to use language needs to evolve in such a system.
Anyway, thats all for now, glad to be back; Any comments (psimagus i know you have one ~lol) on my "ideas", please let me know. tell me why i'm crazy
I must confess, I haven't logged on in almost a year. I've been very busy. But, I'm glad to see that everyone is still here, and that the forge is running on new hardware.
On the note of Artificial Intelligence, here are my latest ideas, for (perhaps Nick 3????)
I am envisioning a system that integrates neural adaptation and evolutionary algorithms. At the lowest level, MLP's (multi-layer perceptrons) will accept input data from cameras and microphones, or any other sensory input. The MLP's will adapt to recognize objects, and output to some form of SOM (self organizing map). the SOM will categorize the data into regions of similarity, which will be stored in memory, which would be a massive heirarchical tree of data. each "Region" of similar data categorized by the SOM will be treated as an object, or an instance of a more theoretical "class" (object oriented programmers are very familiar with that concept). The memory tree will consist of these classes which represent concepts formed from large amounts of input information. This way, after seeing the instances of the same patterns in its perceptions, it can combine all of those patterns into a generalization of them.
On a higher level, a recursive, evolving, self optimizing algoritm takes this conceptual data stored in the tree, and finds patterns and relationships between them. each time the algorithm loops through a data set, it can build on the higer level connections made during its last loop, and finds information and pattern abstractions that could not be found in any previous loops.
At the moment, I am working on a multithreaded neural network model. Lots more work to do.
contrary to earlier projects, I no longer believe that text should be primary sensory input for an AI. Rather, the ability to use language needs to evolve in such a system.
Anyway, thats all for now, glad to be back; Any comments (psimagus i know you have one ~lol) on my "ideas", please let me know. tell me why i'm crazy
It's great to hear from you colonel720. I can't wait to see your ideas in action..I am still having great fun with Nick.
Greetings, Colonel720!
Forgive me for asking, but why 720? I suppose it's factorial 6, but then... why factorial 6?
Forgive me for asking, but why 720? I suppose it's factorial 6, but then... why factorial 6?
I don't remember, i made that as an email address 5 years ago, and pretty much used it for everything else too. i guess the 720 was because i needed something that wasnt taken. The colonel part is a mockery of KFC.
» More new posts: Doghead's Cosmic Bar